This paper presents a design for a microelectromechanical system (MEMS) Multifunctional Reconfigurable Antenna (MRA) capable of tilting its beam in eight directions, along with three new training schemes for mode selection and channel estimation.
Their associated overhead performance and resilience to Channel State Information (CSI) quality degradation are reported. For environments where angles of arrival (AoAs) change relatively slowly, it was found that exhaustive training (ET) was wasteful and that model selection at a reduced rate lessened the training overhead, resulting in an increased system capacity and an impact on bit error rate (BER). On the other hand, for environments in which the AoAs change relatively quickly, a tracking scheme was developed to reduce the impact of imperfect mode selection and channel estimation. Also, the proportion of contribution to system performance from antenna directivity and channel diversity/rank was quantified. Channel diversity/rank was found to have a higher impact on system performance than antenna directivity. In addition, the performance of a random mode selection scheme was studied and analyzed. The random selection algorithm was found to perform just as well as other training schemes in capacity analysis, but it had poor BER performance. 13 figures and 25 references
Report (Grant Sponsored)
Date Published: December 1, 2010